Training a multi-image classification model with a curated vision dataset

In the Training a classification model with a simple curated vision dataset recipe, you went through the steps to ingest a fastai curated dataset and used it to train an image classification model.

In this section, you will go through the same process for another curated dataset called PASCAL_2007. This dataset (described in more detail here: http://host.robots.ox.ac.uk/pascal/VOC/) contains about 5,000 training images and the same number of test images. The dataset includes annotations that identify common objects that appear in each image. The identified objects are from 20 categories, including animals (cow, dog, cat, sheep, and horse), vehicles (boat, bus, train, airplane, bicycle, and car), and other items (person, sofa, bottle, and TV monitor).

The images in the CIFAR dataset introduced in the Training a classification model with a simple curated vision dataset recipe had a single labeled object...